Announcement

Collapse
No announcement yet.

Enermax Liberty ECO 400W [ELT400AWT] - from scraped to serviced

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Enermax Liberty ECO 400W [ELT400AWT] - from scraped to serviced

    Remember this pile of junk electronics I got last year? Well, that's where this Enermax Liberty ECO 400 Watt PSU came from. Like I mentioned, it had cut cables and many bulged caps on the DC output side. I stripped a few cables and did a quick temporary recap with whatever junk (but still reading OK) caps I could find, then performed a basic power test – the PSU passed. Thus, I knew I'd eventually go around to repairing it. One year later and we're here.

    First shown is the label below, so anyone reading can see what PSU we're dealing with here.

    ^ It's model ELT400AWT. Given the strong 3.3V and 5V rail ratings, this is also a fairly out-of-date product… but still viable for a low-mid power modern PC.

    Now the label looks nice and pretty above. But the rest of the PSU did not when I got it. To give you an idea, here are some “before” pictures of what the PSU looked like when I got it (minus the few dust balls I removed - don't like to store dusty hardware):


    As you can see, the fan's missing, output wires are cut, and someone put a “Not Working” label on the case. Clearly it was destined for the scrap heap… which would have been a shame, as Enermax PSUs are designed fairly well. The bad caps aside, I'm pretty sure this PSU can do the power ratings on its label.

    And speaking of which, let's see the rest of the PSU built quality (though I will skip doing a thorough breakdown here like I normally do.)
    https://www.badcaps.net/forum/attach...1&d=1613372320
    https://www.badcaps.net/forum/attach...1&d=1613372320
    https://www.badcaps.net/forum/attach...1&d=1613372320
    https://www.badcaps.net/forum/attach...1&d=1613372320

    ^ Crap caps aside (and cut wires), there isn't that much to complain about here. It's an older group-regulated design with a mag-amp 3.3V rail. If I remember correctly, the 3.3V rail rectifier looked like a 40 Amp –rated part. Same goes for the 12V rail, which was coupled to the rectified DC side of the 5V rail (i.e. 12V rail is a super-imposed 7V rail on top of the rectified 5V rail for better cross-regulation.) And speaking of the 5V rail, I think that one had two 40-Amp rectifiers. So all in all, it's a 5V-heavy PSU also capable of heavy 12V loads (probably depending on transformer output turns ratios and coupled inductor windings.)

    Primary side has APFC with a 400V, 330 uF Hitachi HP cap - good, but not great, given that it's only rated for 400V and 85°C. At least it really is a Japanese brand, like the PSU's product page states. APFC controller is an UCC3817 located on the small daughterboard by the primary cap, along with a LM393 comparator. Meanwhile, the PWM controller is a UC3842bn (actually visible in one of the pictures above.) And the 5VSB also has its own PWM IC, mounted on the solder-side (it's a small 6-pin SMD - SOT-23-6, if I'm not mistaken), but I forgot to look up its numbers.

    Speaking of the solder side, here's a high-resolution picture of that:
    https://www.badcaps.net/forum/attach...1&d=1613372320
    PCB is marked:
    "ENERMAX
    PCB0130-1
    2005/09/29
    Anita Chiu
    "
    Wow! For a 2005 design, this was actually quite good! Back then, PCI-E and SLI were still a new thing, so even some of the most “decked out” SLI PCs would have been satisfied with the power offered by this PSU.

    Back to PSU discussion… this is one thing I never really liked too much about Enermax PSUs – somewhat messy soldering. In particular, here's one thing that really caught my eye while working on this PSU:

    See those 62-Ohm SMD resistors? Those are the dummy-load resistors for the 3.3V rail. And that big blob of solder on the lower-left is nearly shorting the 3.3V rail to ground. It was literally just a flux crust separating the two from shorting out!
    Apart from the excess solder on some joints, the PCB soldering is acceptable.
    Attached Files

    #2
    Enermax Liberty ECO 400W [ELT400AWT] - from scraped to serviced

    Anyways, let’s get to the *fixing* part.

    First I replaced a bunch of the small caps for the logic/controllers on both primary and secondary. I didn’t replace all of them, as I was short on a few values, but I replaced 3x 2.2 uF 50V, 3x 10 uF 50V, and 1x 22 uF 50V - all with a mix of Nichicon PJ and Rubycon YXJ. I think there were two or three more 2.2 and 10 uF total left that I didn’t replace. I also did some part “repositioning” on the primary side - namely, just moved the bridge rectifier and RF choke in front of it so that they could get better airflow.


    Next on the list were the capacitors on the secondary side.

    Starting from left to right…

    First was the 5VSB - here I put a single 1500 uF 16V Nichicon HN cap (out of an Xbox 360 motherboard) for the 1st filter spot, and a 1000 uF 6.3V OST RLP in the 2nd filter spot. The HN may seem counter-intuitive, being ultra-low ESR. But remember, 5VSB uses discontinuous (flyback) design, so caps with very low ESR and ample ripple current handling capability are recommended. I’ve been doing this to a lot of 5VSB circuits now, and so far have not had a problem. Time will tell, of course, if there is too much heat there and if that Nichicon HN can handle it. But I suspect it will be OK. The OST RLP is more like to be problematic and it will be changed at some later point when I do a 2nd revision of the current recap (due to temp recap shown below.)

    Next, we have the -12V rail lone filter cap that goes to the input pin of a 7912 regulator. This cap is a PCE-TUR 25V, 470 uF in 8 mm dia. Unfortunately, I do not have any 25V –rated caps in 8 mm diameter, so I just measured it and left it back in after determining that it is still OK. Again, this will probably be replaced when I get more caps in stock. I do have 16V caps of that capacity and 8 mm dia., but I didn’t measure what voltage the spot gets. Being placed before the 7912 regulator, there is a chance that it could be over 16V (unlikely, but still better to be safe), so I didn’t want to risk it. For those of you recapping this PSU, keep that in mind.

    On that note, there is another thread here on the ELT400AWT. It contains a much more detailed output cap list, so feel free to check it out. In fact, the output caps on my ELT400AWT are exactly the same as the ones listed there.

    With that said, I didn’t have a 3300 uF 16V cap for the 1st 12V rail filter cap, so I used a 2200 uF 16V UCC KYB instead - not ideal, but should be OK for basic testing. My original 3300 uF 16V CTC FM cap actually still measured proper capacitance and ESR. It’s the 2200 uF 16V CTC KM in the 2nd filter spot of the 12V rail that had failed, which I replaced with a recycled Ltec when I was testing if the PSU worked.

    The 5V rail was the same thing: 3300 uF 10V CTC FM cap read normally. However, I replaced it with a 3300 uF 6.3V UCC KY. Again, it’s the 2nd filter cap that had failed (a 2200 uF 10V CTC KM.) I replaced it with a 2200 uF Ltec for testing purposes.

    And finally, the 3.3V rail: it was the same deal as the 5V rail above, except here I used a 3900 uF 6.3V Rubycon ZLQ for the 1st filter cap.

    Clearly this recap job is not complete. But then again, neither is the rest of the PSU. Even after fixing the wires (details noted below), it’s still missing the two PCI-E modular cables, and I’ve only attached a 4-pin 12V CPU cable for the motherboard CPU power (instead of an 8-pin.)

    On that matter, it’s the cut output wires that had me waiting on fixing this unit for so long. In the end, I just took the 24-pin wire harness from a dead/parts Raidmax RX-380K (Sun Pro KY-480ATX). Its wire harness was a joke with ultra-thin 20-AWG wires… but at least they are real copper. Meanwhile, since I had already used the crappy 4-pin 12V CPU connector from that Raidmax on another PSU, I had to come up with something different for the 4-pin 12V CPU connector on this Enermax PSU. Looking through my stash of cables and connectors, I found a really nicely-built female Molex--> male 4-pin 12V CPU adapter with thick 18 AWG wires and sturdy connector pins. But instead of breaking that down and connecting it directly to the cut PSU wires, I soldered a male Molex connector to the PSU with double 18 AWG wires going to the 12V side. In the end, it wasn’t as pretty or well-built as the original design, but it should work. Also, soldering the ATX wire harness to the existing cut wires on the PSU (instead of removing the cut wires) was a good thing, because it made the crappy ATX wire harness from the Raidmax PSU a bit longer. The wires will still likely be too short to use the PSU in a case with a bottom-mounted PSU, but should be OK in a regular older case.

    As a side note, I want to mention that this is one of those instances where my “solder pot” contraption really came in handy. Tinning all those wires in the wire harness one by one with a soldering iron (as well as the cut ones on the PSU) is a slow, painful task - especially trying to do multiple strands twisted into one (due to large thickness and thermal mass.) Basically, the Enermax PSU wires are all 18 AWG and either steel or tin-coated copper, making it difficult to tin with solder and a regular soldering iron. But with the solder pot above, it all became super easy and quick. All I had to do was just put a very tiny bit of flux and dip each strand (or bunch) in the hot solder, count to exactly 2.5 (lol, not really ) and pull out the wire(s) with perfect tinning. No excess solder blobs or partially-tinned wire. Perfect results every time!

    Advertising of my solder pot aside… the last part left was the fan. The ELT400AWL uses a 120 mm fan, which was missing from my unit when I got it. Luckily, my former job was getting rid of a bunch of scrap wires and electronics one day. And from that, I got 3x Yate Loon 120 mm sleeve bearing fans. They were all used and dusty. Two were even starting to grind badly. I took the worst one, cleaned and oiled its bearing properly, and assembled it back together – it was whisper quiet now. These Yate Loon fans also came with both a standard 3-pin fan header and a 4-pin Molex. Of course, since I wanted to utilize the fan controller inside the PSU, I removed the original connector on the PCB inside the PSU and soldered a standard 3-pin fan connector.

    ^ And here you can see that connector soldered to the board and the fan plugged in it. Note that I also put a small label in there to indicate where the (+) and (-) wires go. On a standard 3-pin PC fan, middle red pin is usually positive (+) and the black outer pin is negative (-). But here, I had to have them reversed… which I did on the fan’s connector. While at it, I also removed the Molex connector on the fan, as this just wasn’t going to be needed here.

    Finally, it was time to assemble the unit back together. I ran to the garage and looked through another box of computer stuff that I had picked up a while back from Craigslist, which contained mostly fan guards, ATX extensions cables, some large 200 mm fans, and a bunch of other cables and goodies. In this box, I found a nice 120 mm black fan guard. And for fan screws, I couldn’t find anything big enough to fit the YL fan holes. However, looking through my box of salvaged bolts (many of these came from the floor of a repair shop I used to work at, since no one bothered to ever clean the floors there), I found 4 S.S. bolts with round heads that fit the fan holes perfectly. And lastly, for the PSU top cover - it too was missing its original screws. So what I did is I took the counter-sunk M3-threaded screws from the modular board and used those on the top cover. Meanwhile, since the modular board doesn’t really care about nicely counter-sunk screws, I just used 4 spare M3 screws from my junk box to hold it in place. After all of that hacking and piecing together with random scrap parts I had acquired over the years, this was the result:


    ^ It’s far from complete, but at least it’s functional now. Unfortunately, none of the modular wires were included with the PSU either (and on that note, this is exactly why I hate modular PSUs - people always loose/toss/misplace their modular cables… not to mention modular cables contribute to voltage drop from the extra connectors they have. ) But as my luck would have it again, I dug through yet another box of computer junk that someone on CL gave me after buying a PC case from him… and that box had 3x modular cables with plugs that fit the above Enermax (that is, the one with the 2 blue Molex connectors shown above.) Of course the pin configuration at the connector end was completely wrong, but nothing that a few minutes of extracting pins and placing them in the right slots couldn’t fix. On that note, always check your modular cables when buying used modular PSUs. Some people may be unaware and may include whatever modular cables they find. But different PSU manufacturers sometimes use the same connector, but with different pin configuration. So you could end up putting 5V into a 12V rail, or worse - the opposite! (Or even worse yet, 5/12V into ground and ground into 5V/12V.) So always check this before you power your system.

    Anyways, that’s all I have for this PSU so far. It will need a revision of the recap at some point in the future. But for now, it should be OK to use as yet another test PSU. Moreover, I don’t like the loud high-pitched “electrical arcing” noise it makes when the APFC comes up from stand-by. Given the discussion in this thread, the APFC may not have the best design in these PSUs. So I’m going to use it with caution for now, especially given its age. And due to the lack of PCI-E power connectors currently, it at least won’t see any high loads from powerful GPUs. Though that’s another item I have to sort with this PSU. Anyone has recommendation for cheap modular 6-pin PCI-E cables (or has a few spares?)
    Attached Files
    Last edited by momaka; 02-15-2021, 01:16 AM.

    Comment


      #3
      Re: Enermax Liberty ECO 400W [ELT400AWT] - from scraped to serviced

      Very nice write up and very clean and nice work

      I do have one question for you

      How did the testing of this switching power supply go
      Last edited by sam_sam_sam; 02-15-2021, 08:19 AM.
      9 PC LCD Monitor
      6 LCD Flat Screen TV
      30 Desk Top Switching Power Supply
      10 Battery Charger Switching Power Supply for Power Tool
      6 18v Lithium Battery Power Boards for Tool Battery Packs
      1 XBox 360 Switching Power Supply and M Board
      25 Servo Drives 220/460 3 Phase
      6 De-soldering Station Switching Power Supply 1 Power Supply
      1 Dell Mother Board
      15 Computer Power Supply
      1 HP Printer Supply & Control Board * lighting finished it *


      These two repairs where found with a ESR meter...> Temp at 50*F then at 90*F the ESR reading more than 10%

      1 Over Head Crane Current Sensing Board ( VFD Failure Five Years Later )
      2 Hem Saw Computer Stack Board

      All of these had CAPs POOF
      All of the mosfet that are taken out by bad caps

      Comment


        #4
        Re: Enermax Liberty ECO 400W [ELT400AWT] - from scraped to serviced

        Nice writeup, I would measure the main bulk capacitor, just to make sure the noise is not from it being failed.
        "The one who says it cannot be done should never interrupt the one who is doing it."

        Comment


          #5
          Re: Enermax Liberty ECO 400W [ELT400AWT] - from scraped to serviced

          Originally posted by sam_sam_sam View Post
          How did the testing of this switching power supply go
          The 5VSB is a happy camper with the ultra-low ESR 16V 1500 uF Nichicon HN - no whining or squealing. I load-tested it to only 1.5 Amps, but voltage was fine and didn't sag at all.

          As for the rest of the PSU...
          I haven't done any long or stressful tests with it yet. Only connected it to one faulty motherboard that I know uses about 70 Watts from the 12V rail for the CPU and very little power on the 3.3V and 5V rails, so it's a decent case of a 12V-heavy cross-load. I found that two of my older half-bridge PSUs won't start with it due to the disproportional loading on the 12V rail. But the Liberty ELT400AWT started without problems and 12V rail was just right-on at 12.03V, while the 3.3V and 5V weren't too high (about 3.37V and 5.18V.) Unfortunately, I don't have an oscilloscope, so I can't say what the ripple is like on the output. Though, I suspect everything is OK, since all of the output caps were either checked or replaced with new ones. Of course, I will be getting new caps for it eventually.

          Also a note regarding the 120 mm Yate Loon fan - it's absolutely silent. Sure it's not pushing a lot of air, too... but I can't actually hear it! Maybe that's why I'm over-thinking the APFC is making too much noise - I can hear the PSU's transistors over the fan. This thing is about as quiet as a fanless unit.

          Originally posted by Per Hansson View Post
          I would measure the main bulk capacitor, just to make sure the noise is not from it being failed.
          Yeah, I was intending to do that at last *before* installing the PSU in its case... and I forgot.
          Although the input wires are not soldered and easy to remove, the output wires are a pain - they barely fit through the opening in the case. So once I had it closed, I didn't feel like taking it apart again. There is a possibility that the primary cap could be degraded / starting to fail, but I was like, YOLO at this point.

          It shouldn't be too much of an issue, though, because I still don't have any SATA or PCI-E power connectors, so I can't put this PSU in a power-hungry PC. Thus more than likely, I probably won't use it much (if at all.) Need to get more caps too... but I'm waiting on a few more broken PSUs to show up from eBay to see if I'll need to order anything else. Gotta keep busy with fixing junk, you know.

          On that note, one of them should be a fairly-decent Rosewill Valens 600 unit with 80Plus Bronze or similar. Curious what problem I'll find in that one. The other one is an old-ass Hyper Type-R PSU... because it reminded me too much of FNF, Honda Civics, and V-tech just kicked in, yo!
          Last edited by momaka; 02-15-2021, 10:36 PM.

          Comment


            #6
            Re: Enermax Liberty ECO 400W [ELT400AWT] - from scraped to serviced

            thanks. i really should get down to recappin' mine too, tho mine is the 500w version.

            as for those enermax liberty eternity and infinity modular cables... ah well about that... performance pcs on ebay was having a clearance on those cables at the end of the year in 2019 and i bought a couple from them too on sept 2019 but soon, they've been sold out... there's only one listing i see on ebay for three cables and the cable pouch for US$19.95. thats too much, i guess.

            so i guess u are out of luck. u have to buy the connectors and wires and crimp them yourself. what a fun project for you! since u like doing things hands on and the hard way. hehehe!
            Attached Files

            Comment


              #7
              Re: Enermax Liberty ECO 400W [ELT400AWT] - from scraped to serviced

              okay... not to nitpick, but from the label of the psu posted, this is NOT the liberty eco. the liberty eco is 80 plus standard certified from the label but this one is not. the power ratings are also different. the real liberty eco has less power and current on the minor rails but slightly more power and current on the 12v rail. also, the proper part number for the eco version is ELT400AWT-ECO. this one is just ELT400AWT, so its just the liberty standard non-eco.

              just thought i'd mention that in case newbies on the forum look up repairs for the wrong power supply and it causes issues or explosions!
              Attached Files

              Comment


                #8
                Re: Enermax Liberty ECO 400W [ELT400AWT] - from scraped to serviced

                Originally posted by ChaosLegionnaire View Post
                okay... not to nitpick, but from the label of the psu posted, this is NOT the liberty eco. the liberty eco is 80 plus standard certified from the label but this one is not. the power ratings are also different. the real liberty eco has less power and current on the minor rails but slightly more power and current on the 12v rail. also, the proper part number for the eco version is ELT400AWT-ECO. this one is just ELT400AWT, so its just the liberty standard non-eco.

                just thought i'd mention that in case newbies on the forum look up repairs for the wrong power supply and it causes issues or explosions!
                Good catch!
                I guess too late editing the tittle now.

                However, I do wonder how much more different the "ECO" version might actualy be (if at all) inside from this non-ECO one, then. I don't think the 80Plus certification matters too much, simply because that was started around 2004 but still wasn't a popular standard with many PSU manufacturers. In fact, according to Wikipedia 80Plus page, Seasonic was the first PSU in Feb. of 2005 to use the 80Plus cert. So perhaps this non-ECO version is too old and Enermax just didn't bother to use it back then.

                As for the slightly different current... I don't think that matters too much either. For example, if you look at the label of the 500 Watt version you posted and the 400W I have, the combined current difference on the 12V rail is just a mere 2 Amps. Same with the 5V and 3.3V rails. So that tells me all of these units have a fairly similar configuration for the rectifiers on the secondary side, and perhaps just the primary uses better (or worse) parts, dictating the maximum power the PSU can provide. Thus, the output caps could still use the same configuration for all we know. I guess it would be nice if someone had the ECO version and could post pictures here so we can compare.

                Originally posted by ChaosLegionnaire View Post
                performance pcs on ebay was having a clearance on those cables at the end of the year in 2019 and i bought a couple from them too on sept 2019 but soon, they've been sold out... there's only one listing i see on ebay for three cables and the cable pouch for US$19.95. thats too much, i guess.
                $20 for just a few extra cables?? - Yeah, that is a bit too much! Especially for an old PSU like this that doesn't really have that many extra cables to begin with (only 2x 6-pin PCI-E, which is probably appropriate, given the 12V rail's current capability.) At half of that ($10), I might consider them, though.

                Originally posted by ChaosLegionnaire View Post
                u have to buy the connectors and wires and crimp them yourself. what a fun project for you!
                And most likely that's what I will end up doing... though I might "cheat" a little here - I was thinking of getting a bunch of those cheapo 6-pin PCI-E splitter cables and simply modifying these to make 2x 6-pin PCI-E that the PSU has slots for... or I might go the crimp with new connectors route, if that's cheaper and/or likely to yield better quality connector. IDK yet, will see. Many of the listings for the cheapo 6-pin PCI-E splitters all claim 18 AWG copper wire... but if that claim is like the Ah ratings on cheap rechargeable batteries, we all know how that's going to end:

                Comment


                  #9
                  Re: Enermax Liberty ECO 400W [ELT400AWT] - from scraped to serviced

                  Originally posted by momaka View Post
                  Next, we have the -12V rail lone filter cap that goes to the input pin of a 7912 regulator. This cap is a PCE-TUR 25V, 470 uF in 8 mm dia. Unfortunately, I do not have any 25V –rated caps in 8 mm diameter ...
                  Panasonic FC is/was available in 470uF/25v/8mm, if you still trust it - should be fine for a linear regulator. Even better is NCC KZE in 470uF/25v/8mm, which I stockpiled some 8-9 years ago, foreseeing just this kind of application, especially in SFF PSUs.

                  BTW, there's an NCC KZE 220uF/35v/8 mm, Japanese production in green sleeve, which I stockpiled for audio use about the same time about a decade ago, which also has its uses as a startup cap or similar.

                  I did miss stockpiling a 220uF/16v/6mm KZH when it was available in surplus quantity, though I did take some sample quantity for testing. It's useful for PCI sound card upgrades and so on.

                  Comment


                    #10
                    Enermax Liberty ECO 400W [ELT400AWT] – primary cap check and mods

                    Since people here (and I myself) were curious about the condition of the primary cap, given the “strange” noise the PSU makes when it initially powers up, I decided to pull it apart and check that Hitachi HP3 400V 330 uF primary cap. Besides, I hadn’t touched the soldering iron today, so I needed my daily dose of “the fumes”.

                    Anyways, here’s what the GM328 meter showed:

                    What do you know! – It’s actually still in spec. Capacitance is a bit lower at 295-ish uF, but that’s still within 10-11% of the spec value. So looks like all is well.

                    Of course, since I had the PSU open anyways, I figured this would be a good time to install another mod in there as a test: an extra film or ceramic cap in parallel with the primary input electrolytic cap. A while back, I bought a bunch of 2.2 uF 450V polypropylene metal film capacitors. One was for a replacement of the snubber electrolytic cap in a LiteOn ATX PSU. But the others I got with the intent to install in parallel with the primary caps of various PSUs with APFC circuits. The idea is, at least in my head (haven’t done any sort of research, testing, or measuring to prove this) that the extra ceramic or film cap can take on some of the high-frequency ripple current, thus make things a little less stressful for the primary electrolytic cap (similar to how PC motherboards often have ceramic, tantalum, or other low-ESR and high-ripple current caps in the CPU VRM circuit.)

                    Unfortunately due to limited space on the primary side of the Enermax ELT400AWT, I was not able to fit one of these 2.2 uF caps in there. However, I still wanted to add something in parallel with the primary cap just as a backup, in case it went open-circuit. That way, the APFC coil should still technically have a place to dump its energy and I’m hoping this perhaps could save other APFC components from going into orbit. But who knows.

                    In any case, I have a ton of various salvaged junk boards lying around, so I looked for the highest capacity cap I could find that would fit and meet the voltage requirements of the circuit. The scavenged part came from a Sony CRT TV board: a 2.7 nF 630V 5% film cap of some sort. Now let’s see how I fit it…


                    Basically, I found just enough space between the primary filter cap and APFC coil to squeeze it in there. The PCB already conveniently had a hole on the negative bus trace for a current sense resistor that wasn’t installed on this version of the PSU. But for the positive bus trace, I had to drill my own hole… which I did right in front of the APFC rectifying diode (which actually is a MOSFET with Gate pin soldered to the Source pin and the MOSFET just used as a diode - I guess this is cheaper than using an actual diode? ) After all of this and giving the NTC thermistor a slight push towards the primary heatsink / APFC MOSFET, I was able to fit everything in there and install the primary cap again.

                    So, did this cure the “weird” noise I was hearing with the APFC?
                    Answer: I haven’t tested the PSU yet. It got late and I needed to clean my bench area for tomorrow. As such, this one is… to be continued. However, I doubt this cap alone would change anything. I suspect it’s probably more to sooth my mind than anything else. But perhaps time will tell. I’ll post back when I get a chance to test the PSU again.
                    Attached Files
                    Last edited by momaka; 02-24-2021, 10:58 PM.

                    Comment


                      #11
                      Enermax Liberty 400W [ELT400AWT] - some load tests

                      Originally posted by momaka
                      However, I doubt this cap alone would change anything.
                      And my suspicions turned out to be correct. The APFC still makes the same screechy noise when the main PS is activated. But once the PSU is running, there are no other strange noises and all voltages are OK. In fact, I did some load tests. Read below.

                      First load test consisted of just 6 Amps of load on the 12V rail (~73 Watts, as the 12V read around 12.14V on this test) and slightly over 2.5 Amps on the 5V rail (~13 Watts) with a heating element serving as a resistive bank. I also had a fan cooling the resistive load (about 1.2 Watts draw) and a small incandescent bulb (about 1.3 Watts @ 5V) on the 5VSB rail. More importantly, I attempted this test with a 700 Watt heating element in series with the PSU input line, both to make sure nothing gets damaged if something was to blow up on the primary due to my film cap experiment and to see how the PSU fares with a low line input voltage.

                      Unfortunately, I couldn't get the PSU to start with this setup… but I was actually kind of expecting that. Reason why is because most “Full Range” APFC PSUs are rated to work between 100 and 240 V AC line input. Thus, with the series heating element, the line was dropping below this range. Anticipating this, I had a multimeter measuring the AC line going to the PSU. With the PSU turned off / standby circuit only, the line was sitting right at 120 V AC. I also had a Kill-a-Watt meter connected behind this whole setup (of the PSU and series heating element), confirming the line was ~120-121V. As I jumped PS-ON pin to ground, the PSU fan only twitched for a second and nothing else happened. The multimeter on the AC line confirmed it was dropping a little below 90 V AC, so that explains why the PSU couldn't start.

                      However, I found that if I jumped PS-ON to ground again very quickly after the PSU shut off the first time, I could get it to stay running. So with the above load on the DC output of the PSU (73+13+1.2+1.3 = 88.5 Watts total), my Kill-A-Watt indicated about 130 Watts draw from the whole setup and about 1.09 Amps of current draw (@ 120V AC). Meanwhile, the multimeter measuring the line voltage going to the PSU was showing about 96.5V AC stable with the PSU running. This makes sense: the series 700 Watt heating element (measured approximately 21.7 Ohms resistance) was dropping about 120-96.5 = 23.5V… and 23.5V / 21.7 Ohms = 1.083 Amps, which agrees with what the Kill-a-Watt was indicating.

                      Therefore, the PSU must be drawing about 96.5V x 1.083A = 104.5 Watts on the primary side. Dividing the output power by the input yields 88.5 / 104.5 = 84.7% efficiency according to my calculations… and that's with a lowly 96.5V AC line – not bad! (That is, if my calculations are correct… which I think they are in this case.)

                      Anyways, I repeated this test once more, but with a 1400 Watt series heating element so that I wouldn't have to try to “trick” the PSU to start. And sure enough, it did start fine. I don't remember what numbers I got for this test, as I didn't write them down, but the efficiency was mostly the same - maybe a fraction of a % higher.

                      Then I tried another test: switched back to the 700 Watt series element, but connected another resistive bank to the 12V rail for an additional 6 Amps more. I expected the PSU not to be able to start, as the AC input line was already below the 100-240 V AC range with just the single 6 Amp load on the 12V rail. Sure enough, when I jumped PS-ON to ground, the APFC screeched for a brief moment, the PSU fan twitched a little, and then the PSU did NOT turn On. This time, no amount of “tricking” with cycling the PS-ON signal would make the PSU run… which is GOOD! ATX power supplies are not designed (and should NOT try) to keep running in brown-out conditions. So the Enermax ELT400AWT here was correctly shutting down due to an abnormally-low AC line. … unlike the KDMPower MIPC MI-X8775CD PSU I tested a few weeks ago, which wanted to just keep going on any line voltage.

                      After these tests, I removed the series heating element and plugged the PSU straight in the wall. Now that there was nothing to drop the AC input line, I could load the PSU as high as I wanted… or as high as my resistive load bank allowed me to. On that note, I'm using a Ni-Cr heating element (rated for 3000 Watts and 240V AC) from an electric clothes dryer as the resistive bank. My limitation is not so much this element, but the number of power connectors going to it (currently 1x 4-pin 12V CPU, 1x 6-pin PCI-E, and a main 20-pin ATX), which limits how much current I can pull from the 12V rail with the wiring I have on these. With that said, I was able to pull up to 3x 6 Amp (~18 Amps total) from the 12V rail, with the 5V rail still being loaded only with ~2.5 Amps and the 3.3V rail unloaded. You might expect this heavy 12V cross-load to give poor regulation from the PSU… but that wasn't the case here. The 12V rail was sitting at 11.97V, which while below 12V, is still pretty darn close to spec. Meanwhile, the 5V rail didn't go higher than 5.15V and the 3.3V rail was doing well at 3.38V. Good job Enermax! For a group-regulated PSU, this one surprised me in a good way. Of course, the above load on the 12V rail is only slightly more than half of what this PSU is capable of, so things could change with a higher load (to be tested another time when I have my load bank setup for higher currents.) Also, obviously it's a shame I don't have an oscilloscope to check the ripple output. But I'm pretty sure it's still in spec with these loads, despite the slightly smaller cap I installed on the 1st filter spot of the 12V rail.

                      So aside from waiting for a more proper recap and finding/making the other (missing) modular cables, this PSU is more or less complete and ready to work. As for the screeching noise from the APFC on start-up, I guess that's just the nature of the beast. I've seen other APFC PSUs make that noise too, though just not as loudly. The addition of that 2.7 nF film cap I put in parallel with the primary electrolytic cap didn't change anything in terms of noise. However, I noticed that when I had the PSU running with the series 700 Watt heating element (the test where I had to “trick” the PSU to start), the APFC did not make any noise at all. I suspect this is due to the resistive heating element in series with the input. When I switched it for the 1400 Watt heating element, the screeching noise from the APFC was audible again. And finally, when I had the PSU running directly from the wall, the screeching noise from the APFC was the loudest. So perhaps it just has to do with the design of the APFC circuit in this PSU. Time will tell, I suppose, if the APFC takes out my primary cap… and if that small film cap I put matters. In the meantime, I'll probably use this PSU now as-is, since at least I know it can provide up to 18 Amps on the 12V rail, which is enough for many of the systems I deal with around here.

                      Comment


                        #12
                        Re: Enermax Liberty ECO 400W [ELT400AWT] – primary cap check and mods

                        Originally posted by momaka View Post
                        2.2 uF 450V polypropylene metal film capacitor. a 2.7 nF 630V 5% film cap. I still wanted to add something in parallel with the primary cap just as a backup, in case it went open-circuit. That way, the APFC coil should still technically have a place to dump its energy and I’m hoping this perhaps could save other APFC components from going into orbit. But who knows.
                        im still quite sceptical of that backup primary cap actually working and preventing a blowout of the apfc section if things go wrong. 2.2uF is 2200nF and 2200nF vs 2.7nF is quite a big difference. im not sure if 2.7nF is enough capacitance to hold the excess energy from the apfc coil to prevent a blowout especially when we're talking about high apfc boost voltages of 390V or more. not sure what the math formula is to calculate how much electrical energy in joules a capacitor can hold based on its capacitance and working voltage but i think it helps to do your math on paper in theory beforehand...
                        Originally posted by momaka View Post
                        But for the positive bus trace, I had to drill my own hole…
                        lol! i have a lot to say about the double meaning of that but i shall save the dirty talk and jokes for the vip room... hahaha!
                        Originally posted by momaka View Post
                        which I did right in front of the APFC rectifying diode (which actually is a MOSFET with Gate pin soldered to the Source pin and the MOSFET just used as a diode - I guess this is cheaper than using an actual diode?)
                        yeah who knows? maybe they were cheapskates or it could be that it was done that way because its a trademarked, patented and trade secret design that serves some secret purpose only the designers know of that we ordinary folk arent aware of? hahaha!

                        anyway, my hypothesis here is that they used that particular mosfet that way because it has better power efficiency than using a regular diode (lower forward voltage drop perhaps?) but im a nub when it comes to apfc and diodes so i could be wrong... so dont quote me on that...
                        Originally posted by momaka View Post
                        84.7% efficiency according to my calculations… and that’s with a lowly 96.5V AC line – not bad! You might expect this heavy 12V cross-load to give poor regulation from the PSU… but that wasn’t the case here. Good job Enermax! For a group-regulated PSU, this one surprised me in a good way.
                        oh i actually already knew all of this because i read a ton of reviews about the enermax liberty line before buying it in 2008. when it first came out in 2006 or 2007 or so, it was the best power supply to get for your computer back then. many reviews were raving about its high efficiency back then. i recall one review said it had 75% efficiency at max load which was quite impressive considering the junk psus back then in 2006/7. since u were using it at low to moderate load plus with depressed line voltage, the actual efficiency at proper line voltage should obviously be higher and its around 80 plus silver territory at low to medium loads. very nice! in my uk/eu 240v line voltage land, i'd dare say the efficiency would be higher at 80 plus gold territory!

                        i also recall another review that crossloaded this psu and they said it had no crossloading blind spots where crossloading the psu produced voltages that were out of atx specs, so this psu can be crossloaded with a wide variety of load conditions. minor rail heavy p3 cpus and gpus and 12v rail heavy p4, athlon 64 cpus and gpus are no problem for it. therefore, this is a very flexible psu to use with whatever system u please.

                        so thanks for posting this analysis and study of this psu anyway, even though i am already aware of all of this... u just reminded me of how wonderful this psu was back then and u now make me wanna hoard all of the enermax liberty psus i can get my hands on on junkbay! hahaha!
                        Originally posted by momaka View Post
                        it’s a shame I don’t have an oscilloscope to check the ripple output. But I’m pretty sure it’s still in spec with these loads, despite the slightly smaller cap I installed on the 1st filter spot of the 12V rail.
                        i think gabriel torres on hardware secrets also scoped the psu outputs in his review. it had ultra-low ripple (15mV) on the outputs as well at 80% load which gives the superflower leadex platform a run for its money since its a 2015-ish platform while the enermax liberty platform is from nearly a decade ago from 2006. plus the superflower leadex platform used japcaps to get its ultra-low ripple while the enermax liberty used taiwanese second/third tier caps to get its ultra low ripple. so when properly recapped with 10k hour japcaps, the enermax liberty should be able to hold its own against all the modern 10 year warranty, 80 plus gold psus currently out there and maybe even outlive some of them with the new 10k hour japcaps u put in. who says good ol' stuff isnt better than "modern junk"?
                        Last edited by ChaosLegionnaire; 03-15-2021, 08:28 PM.

                        Comment


                          #13
                          Re: Enermax Liberty ECO 400W [ELT400AWT] – primary cap check and mods

                          Originally posted by ChaosLegionnaire View Post
                          im still quite sceptical of that backup primary cap actually working and preventing a blowout of the apfc section if things go wrong. 2.2uF is 2200nF and 2200nF vs 2.7nF is quite a big difference. im not sure if 2.7nF is enough capacitance to hold the excess energy from the apfc coil to prevent a blowout especially when we're talking about high apfc boost voltages of 390V or more.
                          Well, this Enermax Pro 82+ 525W has both its primary caps go open, and the only thing left to absorb energy on the primary was a 0.1 uF ceramic cap (i.e. 100 nF). Nothing blew, aside from the NTC going high-resistance. Granted 2.7 nF is almost 33 times lower capacitance than that, but I still think it might help vs. having nothing in there, in which case the voltage from the APFC coil can really go sky-high until it finds a path to go through. So even a little bit of capacitance can make a lot of difference vs. none at all. Indeed, of course, the 2.2 uF cap would have been better. But I might just order a 1 uF cap on the next cap order I do and swap that in there. That's still 10 times more capacitance than the ceramic cap in the Pro 82+ 525W.

                          Now, how much energy the APFC coil holds per a single pulse... that depends on its inductance and average current. I forgot what the formula is here and too lazy to go back through my analog power circuits books. We did briefly touch on boost and buck converters at one point, though, and I remember it wasn't too complicated. But it's not something I've had to use too often, so it expired from my brain.

                          Originally posted by ChaosLegionnaire View Post
                          not sure what the math formula is to calculate how much electrical energy in joules a capacitor can hold based on its capacitance and working voltage but i think it helps to do your math on paper in theory beforehand...
                          E = 0.5 * C * (V^2)
                          - where C is in Farads and V is in volts.

                          So for a 330 uF cap charged to 390V...
                          E = 0.5 * (330 * 10^-6) * (390^2) = 29.0965 Joules of energy

                          2.2 uF...
                          E = 0.16731 Joules, or less than 0.6% of the energy in the primary cap

                          That means the 2.7 nF cap with its capacitance being roughly 1000 times smaller than the 2.2 uF, it can hold only roughly 0.0006% of the energy of the big cap. However, if we assume the primary cap goes open and let the voltage jump all the way up to 550V (very high, but still not high enough to break down any other MOSFETs or transistors on the primary side), then that figure changes to 0.0014%. That's still very small... but again, better than a complete open circuit. So this might buy just enough time for the APFC controller to see the OV condition and stop boosting the APFC... I'm hoping!

                          Yeah, I need to do more calculations, but I'm lazy ATM. So instead, I turn to experiments similar to Garage54, except with electronics rather than cars (read: Ladas ).

                          Originally posted by ChaosLegionnaire View Post
                          anyway, my hypothesis here is that they used that particular mosfet that way because it has better power efficiency than using a regular diode (lower forward voltage drop perhaps?) but im a nub when it comes to apfc and diodes so i could be wrong... so dont quote me on that...
                          Well, I'm not quoting you to "quote you", but rather to reply.
                          We could figure it out if we find the datasheet of that MOSFET and see how much it costs at a retailer vs. an FR HV diode and then also compared their voltage drops... but that's just too much work. So I will assume the truth is somewhere in between: that is, the MOSFET is cheaper and also provides "adequate" performance when it comes to Vf vs. something more expensive/exotic.

                          Originally posted by ChaosLegionnaire View Post
                          so thanks for posting this analysis and study of this psu anyway, even though i am already aware of all of this...
                          You're welcome!

                          Originally posted by ChaosLegionnaire View Post
                          so when properly recapped with 10k hour japcaps, the enermax liberty should be able to hold its own against all the modern 10 year warranty, 80 plus gold psus currently out there and maybe even outlive some of them with the new 10k hour japcaps u put in.
                          Maybe.
                          I'm still putting my money on the APFC cap blowing before anything else. So 10k-hour Japanese caps or not, I don't think this PSU will ever be as reliable as a non-APFC PSU. On that note, I did buy something very "interesting" last month. Pictures to come probably a little later this month in the PSU build thread (or wherever.) All I'm going to reveal about it is that it has a pink-ish paint job from the factory!
                          Last edited by momaka; 03-19-2021, 10:09 PM.

                          Comment

                          Working...
                          X